Proposes an adaptation method for a behavior-based locomotion robot. Utilization of a behavior-based controller makes the controller designing process easier and shorter, because the designer can deal with behavior controllers for simple behaviors independently, and after that can coordinate those behavior controllers in order for a robot to achieve the objective complex behavior. Some problems are still remaining. One is how to adjust the total behavior when the target task or environment is changed. We propose the method to adjust the behavior coordinator against some changes. This method adjusts the activation level of each behavior controller for the changes in fewer trials. It is applied to real brachiation robot control. The brachiation robot has a redundant mechanism to locomote from branch to branch like a long-armed ape.

The authors propose an adaptation and learning algorithm to adjust behavior coordinator against small changes of the objective task or a new one. This method measures the relation between the each local behavior and the global behavior, and then determines the direction of change of the activation values from the behavior coordinator based on the measurements. The monkey-type locomotion robot, Brachiator III has 13 links and 12 joints and is able move like a real ape in a three-dimensional space. The controller is designed for a primitive case. The proposed algorithm is used when the branch interval is changed and when continuous locomotion is desired. We show the effectiveness of the proposed algorithm through experiments using Brachiator III.